For investors trying to time the market, spotting a bubble is hard enough. Predicting when it will burst is nearly impossible, as the legends of Wall Street discovered during the dotcom mania. Today, with AI stocks soaring and valuations reaching dizzying heights, the same fears are resurfacing. Could unconventional indicators, like a surge in public interest, provide a clearer warning signal than traditional metrics?
The Dotcom Lesson: Legendary Investors Got It Right, But Too Early
In the late 1990s, financial titans saw the dotcom bubble forming. Ray Dalio of Bridgewater warned of a "blow-off phase" in 1995. Peter Lynch of Fidelity noted a lack of worry among investors that same year. By 1996, Howard Marks was concerned that everyone from cocktail party guests to cab drivers was discussing hot stocks.
Yet, their early calls came at a cost. George Soros lost $700 million by shorting internet stocks too soon in 1999. Warren Buffett's Berkshire Hathaway underperformed the soaring NASDAQ by an average of 15 percentage points annually in the five years leading up to 1999. The index skyrocketed by nearly 1,100% between 1995 and its peak in March 2000, before crashing by over 80% in the next two and a half years. The lesson was brutal: identifying a bubble is far easier than timing its collapse.
Why Traditional Valuation Measures Fall Short
Today, investors eyeing the stratospheric rise of AI-related companies are asking the same questions. Stocks like Palantir trade at over 200 times expected earnings. The S&P 500's valuation, relative to underlying sales, is more than 60% higher than it was at the peak of the dotcom boom.
However, traditional tools like the Cyclically Adjusted Price-Earnings (CAPE) ratio are of little help for short-term predictions. While high CAPE readings strongly correlate with low returns over a ten-year horizon, they show no correlation at all with performance over a single year. This leaves investors searching for better, real-time indicators of market euphoria.
The Google Search Signal: A Modern Bubble Gauge
Following Howard Marks's cocktail-party test, one novel approach is to track public frenzy through Google search trends. The logic is simple: a bubble often attracts a flood of retail investors at its most precarious peak, just before it pops.
Analysis of recent manias shows a telling pattern. Significant spikes in searches for terms like "bitcoin," "Dogecoin," "SPACs," and Cathie Wood's "ARKK" fund consistently preceded sharp price declines over the following 12 months. For several, including GameStop and SPACs, the peak in search interest coincided almost exactly with the peak in price.
This pattern is not foolproof. Searches for "AI stocks" hit their zenith in mid-August 2025, yet prices continued to climb for weeks. However, it is notable that search interest has since plummeted dramatically, coinciding with recent market wobbles. Nvidia's share price fell 15% from its peak, the Philadelphia semiconductor index dropped 13% in November, and the AIQ ETF saw a 12% peak-to-trough fall.
Another Ominous Sign: When the Bears Throw in the Towel
History offers another non-traditional signal: the demise of prominent bearish funds. During the dotcom bubble, legendary investor Julian Robertson was forced to close his fund on March 30, 2000, after years of avoiding tech stocks and facing massive client withdrawals. Ironically, the NASDAQ had peaked and begun its collapse just two days earlier.
In a potentially parallel move, Michael Burry – famous for betting against mortgage-backed securities before the 2008 crisis – informed investors in late October 2025 that he was closing his fund. This year, Burry had been actively shorting AI stocks, including Palantir and Nvidia. The closure of such a conspicuously bearish fund could be a contrarian indicator worth watching for those trying to time the current market cycle.
Ultimately, the data suggests that while valuations tell a story of long-term risk, the fever pitch of public interest – measured by digital footprints – and the fate of high-profile skeptics may offer more timely clues about when a bubble is running out of air.
